Gates Package — page 1353

® PREDICT AND DESIGN DE NOVO ODORS @) HOW ARE BEHAVIOR-MODIFYING ODORANTS IDENTIFIED? KEY QUESTION(S) -------- : I oe ae aaa Scent prediction Can we accelerate the discovery of novel Trained neural network odors using Al and ML? [> -- for prediction > Floral nuit Meaty Lavender Savory \ FUTURE STATE Molecule Nodes © --illl 138 odor descriptions Jasmine Sealy Rapid identification of entirely new Gia Ly cis ceomy classes of chemicals and odors that have mela Laie esc wom impact on human health and experience network Lj ce seeriok: bes Cognac Roasted layers uy Fermented Winey Novel molecule Wr Seg lea Alcoiclic: discovery for behavior POSSIBLE APPLICATION(S) Graph I a) ae 4 New classes of pest repellants that to vector x improving taste of protect against vector-borne infections operation [IEIT Ti-- food) or attractants that enhance crop health © WHAT PARTNERS/INCENTIVES COULD ENABLE THIS WORK? -- PARTNERS: INCENTIVES: Fast-tracked approval pathways for odorant-controlled TROPIG since GS sCJohnson USDA behaviors, costing structures for low- and middle-income contexts (to be developed) G if Not for distribution